Investigating the Effects of Spectroscopic Method in Estimating Soluble Solid Content Values and Firmness of Cherries from an Environmental Point of View: Prediction of Environmental Parameters with Machine Learning Method

In this study, 60 cherry samples with native varieties were selected from the Hir region in Ardabil province. They were classified into four growth stages, including before the optimal harvest date, the day before the optimal harvest time, the optimal harvest time, and after the optimal harvest date, by a panel of human experts. Next, by combining the feature selection method (relief) and the spectrometry method (vis-NIR), the effective wavelengths were extracted to estimate the soluble solid content (SSC) values and firmness of the cherry product. In the continuation of the process of this method, a list of inputs was formed, and by applying the life cycle assessment method, the environmental effects of the process of estimating SSC values and cherry hardness in the presence of tests and obtained data was performed. In the final stage, with the help of the radial basis function neural network method, a relationship was established between the reflection intensity values in the effective wavelengths and the endpoint effects of the life cycle assessment to estimate the environmental effects. It was found that the radial basis function neural network could estimate the environmental effects of the experimental process with an acceptable accuracy (over 95% on average).